Prioritized Path Planning of Multiple Autonomous Vehicles in Urban Environments

Autonomous Vehicles (AVs) are poised to bring in the next transformation in urban mobility. However, for efficient performance these vehicles need to be able to leverage the information gathered from the environment to plan optimal paths to their destination. Simultaneous planning the paths of multiple AVs is a challenging task because of the stringent safety and efficiency constraints. Prioritized approaches provide a practical solution to this problem by decoupling the simultaneous path planning problem into a series of individual path planning tasks based on a priority order. In this paper we present a new prioritized approach based on the PD∗algorithm for the path planning of a system of AVs in a given environment. The PD∗algorithm has been extended to incorporate multiple destinations, delayed starts and a new prioritization metric based on a combination of freedom-index and distance to target; to make the algorithm more suitable to the unique challenges of AV path planning. Further, the suitability of the algorithm has been demonstrated through simulations on the Moving AI Lab's database of city maps in multiple scenarios.